Challenge description:

Optical Coherence Tomography (OCT) is a retina noninvasive imaging technique widely used for diagnosis and treatment of many eye-related diseases. Different anomalies such as Age related Macular Degeneration (AMD), diabetic retinopathy or Diabetic Macula Edema (DME) can be diagnosed by OCT images.

Due to the importance of early stage and accurate diagnosis of eye-related diseases, providing high resolution and clear OCT images is of high importance. Therefore, analyzing and processing of OCT images have been known as one of the important and applicable biomedical image processing research areas.

Different processings have been applied on OCT images, such as super-resolution, de-noising, reconstruction, classification and segmentation. Despite many algorithms working on OCT image analysis, still there is a need for improving the quality of the resulting images and the accuracy of classification.

In this challenge, we mainly focus on the following three problems

1. De-noising of a noisy OCT dataset.
2. classification of a OCT dataset into several sub-classes.  The aim of this challenge is to classify the provided OCT images into Healthy, diabetic patients with DME and non-Diabetic patients.

The  dataset includes 161 raw and noisy OCT images each includes 300 B-scans of size 300×300  pixels.  The 80% of the dataset including 129 patients will be released for train and 32 patients will be kept for test.   The test dataset will be released in an announced time.  The participants are asked to classify the dataset into Healthy (0), Diabetic (1) and non-Diabetic patients (2). The complete description of the dataset can be found in next step and also our website: https://misp.mui.ac.ir/en/ss-oct-image-classification/

Our recently published paper using this dataset, has been mentioned in the reference section.

Approval of all ethical and experimental procedures and protocols was granted by the Institutional Review Board and Ethics Committees of the National Institute for Medical Research Development under Approval Nos. IR.MUI.RESEARCH.REC.1398.732, IR.MUI.RESEARCH.REC.1398.803, and IR.NIMAD.REC.1397.314.

Keywords: Optical Coherence Tomography (OCT), Classification

Data

The participants included people with DME or “1”, Healthy or “0”, and Non-diabetic patients or “2” with AMD, CNV, or Macular Hole (MH) disorders. Each subject contains 300 B-scans with 300 to 1200 x 300-pixel resolution.

Since some of the data have poor quality or the macular region was not captured, Quality Assessor (QA) was designed to automatically determine if the B-scan is suitable for further analyses or not. For the data preparation step, the manual quality assessment was performed on 191 volumes.

Raw Dataset: Includes 161 volumes which has been obtained after image quality assessment with automatic cropping. Each subject contains 30 to 300 B-scan with 300 x 300-pixel resolution. The participants included people with DME or “1”, Healthy or “0”, and Non-diabetic patients or “2” with AMD, CNV, or Macular Hole (MH) disorders.